Underestimation of the early risk of recurrent stroke: evidence of the need for a standard definition.

BACKGROUND: There is considerable variation in the definitions used for recurrent stroke. Most epidemiological studies exclude events within the first 28 days (eg, Monitoring Trends and Determinants in Cardiovascular Disease [MONICA]) or events within 21 days in the same territory as the presenting event (eg, most stroke incidence studies). However, recurrence is most common during this early period and these restrictive definitions could underestimate the benefits of early prevention.
METHODS: We determined the 90-day risk of recurrence after incident ischemic stroke in 2 population-based cohorts (Oxford Vascular Study [OXVASC] and Oxfordshire Community Stroke Project [OCSP]) with the 3 most common definitions: any stroke > or =24 hours after the incident event excluding early deterioration not caused by a stroke (definition A); as above, but excluding any stroke within 21 days in the same territory as the incident event (definition B); and any stroke > or =28 days after the incident event (definition C).
RESULTS: 657 patients had 93 recurrent strokes between 24 hours and 90 days after the incident event. The 90-day recurrence risks (95% CI) using definition A were 14.5% (11.5 to 17.5) in the OCSP and 18.3% (10.8 to 25.8) in the OXVASC. The equivalent risks using definitions B and C were 8.3% (5.9 to 10.8) and 4.8% (2.8 to 6.7), respectively, in the OCSP and 7.0% (1.6 to 12.4) and 5.9% (1.0 to 10.9) in the OXVASC. The definition A risk of recurrence was particularly high after partial anterior (22.9%,17.5 to 28.2) and posterior (19.5%,13.0 to 25.9) circulation strokes.
CONCLUSIONS: The 3 most widely used definitions of recurrent stroke yield markedly different 90-day risks. We suggest that, where possible, definition A be adopted as the standard to avoid underestimation of risk and to allow valid comparison of different studies.